What is the origin of the crappy singleplayer RTS campaign?

34thcell

Look, Ma! Two Heads!
So I realised - many years ago - that most RTS games (of those that I've played) are multiplayer games with crappy singleplayer campaigns attatched, and I merely took this as a given. So now I am wondering what the origins of the crappy singleplayer campaign are i.e. AOE, AOM, SC.
 
Ehh its not games fault its AI...play Warcraft 3 or Starcraft online you will have lots of fun...

Any way if you want rts try older games as funny as it may sound they kinda better in terms of game play...
 
I think starcraft had a decent single player campaign, well written and intresting.
 
Dragula said:
I think starcraft had a decent single player campaign, well written and intresting.

Yes but AI was not a chalange right? actually i enjoyed SC campaign only becouse it was like you said well written and interesting...
 
C&C Generals, Red Alert 2, Starcraft and Warcraft 3 all had really fun campaigns.

I can't think of any that I've played recently that weren't fun.
 
X-Com 1&2,Dune Battle for Emperor,Settler 2,3,4.Europa Universalis III,Hearts Of Irons II there are many good games but well kids like shiny stuff so now we got dumb RTS games with lots of BLOOM and BOOOM :)
 
You're pretty much going to hear AI repeated over and over.
If you're going to compare titles, there is such a thing as good AI in a strategy game, but in general, without drawing comparisons, there isn't a single strategy game in existence with decent AI in all actuality, an AI that can react fluidly to different situations and at the same time not be an incredible dweeb who makes inexcusable mistakes.
I'm pretty sure this goes twice as strongly for TBS games, but that isn't the topic.

@gregor_y - He was talking about RTS games, you mentioned like a single RTS title and one slow-paced road management series (I do like Settlers, but I'm just saying that most of the time that's all it really is, or maybe they changed it post Settlers 2, I haven't played the later ones).
 
I don't remember a RTS that had a decent AI . . . They're all pretty manipulative, or just cheat to get the advantage.

I mean, the RTS combat of Lord of the Realms 2, you just spammed the enemy with, say, maces and you would win any battle. In C&C, for some reason the enemy never noticed the giant mass of infantry prancing towards their base (if only they'd use grenadiers! Or use their tanks to squish you into a gelatinous blob) . . . Now, War Wind (an SSI title from 1996) had a pretty good AI (that was horrible at building bases). . . But that's so long ago now, I might be completely wrong. Hell, X-Com: Apocalypse has AI equal to Star Craft's . . . Down right stupid. Warcraft. Warcraft had decent AI, no?

Only good AI I know of is in turn-based: X-Com and Sierra's Civil War Generals 2, Shogun: Total War, and Sid Meir's Civilization on high settings.
 
JayGrey said:
I don't remember a RTS that had a decent AI . . . They're all pretty manipulative, or just cheat to get the advantage.

I mean, the RTS combat of Lord of the Realms 2, you just spammed the enemy with, say, maces and you would win any battle. In C&C, for some reason the enemy never noticed the giant mass of infantry prancing towards their base (if only they'd use grenadiers! Or use their tanks to squish you into a gelatinous blob) . . . Now, War Wind (an SSI title from 1996) had a pretty good AI (that was horrible at building bases). . . But that's so long ago now, I might be completely wrong. Hell, X-Com: Apocalypse has AI equal to Star Craft's . . . Down right stupid. Warcraft. Warcraft had decent AI, no?

Only good AI I know of is in turn-based: X-Com and Sierra's Civil War Generals 2, Shogun: Total War, and Sid Meir's Civilization on high settings.
Good observation.

It is much, much easier to write a good turn-based AI than it is to write a good real-time AI.

It isn't really a lack of attention to AI, it's mostly a case of difficulty.
 
Wait! Wait! There is one game I played, a RTS with an excellent AI . . . But it wasn't quite a AI. It was made by the same people who created the Commandos WW2 series, I believe. Isometric, with hand-drawn landscapes, and that sort of thing.

Trouble was, everything was scripted.

Say you're the Allies. Well, in Mission X you must stop a evil Nazi convey from gettin' by. This much is scripted -- You know where they're going to be, and when. So, you got about 10 minutes to plan an ambush with mines and infantry.

So, enemy drives along . . . If they spot your men, they haul their asses off the road and open fire. If they hit a mine, they stop and go defensive .etc

But, it's not a real AI: It's scripted actions.
 
JayGrey said:
Wait! Wait! There is one game I played, a RTS with an excellent AI . . . But it wasn't quite a AI. It was made by the same people who created the Commandos WW2 series, I believe. Isometric, with hand-drawn landscapes, and that sort of thing.

Trouble was, everything was scripted.

Say you're the Allies. Well, in Mission X you must stop a evil Nazi convey from gettin' by. This much is scripted -- You know where they're going to be, and when. So, you got about 10 minutes to plan an ambush with mines and infantry.

So, enemy drives along . . . If they spot your men, they haul their asses off the road and open fire. If they hit a mine, they stop and go defensive .etc

But, it's not a real AI: It's scripted actions.
This is one of the ways around the constraints of a real-time AI. This doesn't work for more complicated games, though, and certainly not for the games that are great multiplayer games, which have to be reasonably complex to present a strategic challenge.

Other ways are AI cheating (full map knowledge, for instance), and having a limited number of fixed strategies (for instance, build orders).

All of these approaches have drawbacks, though.
 
Sander said:
JayGrey said:
I don't remember a RTS that had a decent AI . . . They're all pretty manipulative, or just cheat to get the advantage.

I mean, the RTS combat of Lord of the Realms 2, you just spammed the enemy with, say, maces and you would win any battle. In C&C, for some reason the enemy never noticed the giant mass of infantry prancing towards their base (if only they'd use grenadiers! Or use their tanks to squish you into a gelatinous blob) . . . Now, War Wind (an SSI title from 1996) had a pretty good AI (that was horrible at building bases). . . But that's so long ago now, I might be completely wrong. Hell, X-Com: Apocalypse has AI equal to Star Craft's . . . Down right stupid. Warcraft. Warcraft had decent AI, no?

Only good AI I know of is in turn-based: X-Com and Sierra's Civil War Generals 2, Shogun: Total War, and Sid Meir's Civilization on high settings.
Good observation.

It is much, much easier to write a good turn-based AI than it is to write a good real-time AI.

It isn't really a lack of attention to AI, it's mostly a case of difficulty.

Are you kidding me? I haven't played a single 4X game with AI that didn't make me want to dispose of the game after awhile, most turn based games are ruined by their AI.
I could at least stand RTS AI simply because the games travel at such a pace that major AI foibles aren't really that noticeable, and if they did do terribly stupid things they typically made up for it by utilizing their ability to build units faster than you since they didn't have to set up hotkeys or anything to access their production centers.
In an RTS game all the AI has to do is maintain the pace, produce a half-way decent selection of units, at least attempt to counter what you have while not leaving their sides open.

Kohan: Immortal Sovereigns had great AI (at least comparatively) and CoH's AI was reactive and kept its holes covered up (it was garbage otherwise) and both of them generally satisfied on the counts that mattered for OK AI.

And whoever said Civilization is a liar, the AI is both a cheater and a latent pacifist butthole, I don't think the AI in Civilization IV even has wars amongst each other, and if they do it probably lasts ages since all they tend to do is stack hundreds of units on top of each other in every single one of their cities.
Also spamming specific units in most TBS games is the order of the day, you can go on about proper unit counters and how the TBS system provides more flexibility in the realm of unit balance, but in the end it's the person who has the more advanced technology and is more capable of pumping that technology out in weapon form who wins.

This is especially freaking annoying in Master of Orion II when on the harder settings the AI is still so slow that I eventually decimate them before they even have enough pew pew lasers to destroy a single one of my ships.
My belief is this. TBS games tend to be so complex, and tend to have far more possibilities than RTS AI because of that, I mean look at the various strategies, possible tactics, and interactions that Civ IV holds and compare it to Starcraft or CoH, typically in those games units move towards each other, shoot, and occasionally use special abilities. If you're playing online, there's some microing, and even CoH's singleplayer has the AI manuevering its tanks, but that's really about it.
In a TBS game the AI simply has more chances to fuck up, and fuck up majorly, it's not like an RTS where these mistakes typically don't mean all that much because the AI immediately makes up for it with immediacy.

The thing people have to be happy for is that the AI is at the very least adequate for the most part.
On one side of the spectrum you have Galactic Civilizations II which probably has the best turn based AI around, and that isn't saying much, and on the other there's Ascendancy, which is bogglingly idiotic.
 
So ehm, where, exactly do you disagree with my statement that a real-time AI is harder to write than a turn-based AI?

The fact that they haven't been able to approach human intelligence in either isn't exactly surprising (they've barely managed to do that in chess, and that's a pretty simple and well-defined game).

Look, the problem with AI scripting in any reasonably complex game is that (without AI cheating) you can't really get further than a simplified version of what the AI writers are personally capable of..

Eyenixon said:
And whoever said Civilization is a liar, the AI is both a cheater and a latent pacifist butthole, I don't think the AI in Civilization IV even has wars amongst each other, and if they do it probably lasts ages since all they tend to do is stack hundreds of units on top of each other in every single one of their cities.
I've seen plenty of intra-AI wars in CIV 4, especially with the aggressive AI, so I have no idea what you're blabbing on about. Hell, I've repeatedly had it happen that by the time I got to a new continent, one of the civilizations there had already wiped out another one.

But you also have to keep in mind that the games are also built with enjoyment in mind. The AI creators of CIV 4 have actually discussed this at times, and if I recall correctly a very warlike AI mostly annoys a lot of players who wish to build an empire and sucked the enjoyment out of the game. Which is why they split off the Aggressive AI.

Also, from what I know Civilization 4 AI on the high levels is considered to be a real challenge.
 
I've never seen them have wars with each other, they typically sit on their ass and do nothing.

Also I was making the point that TBS AI is traditionally atrocious, it's harder to make AI for TBS games simply because there's so much more that has to be accounted for, RTS AI is typically easier because in the end the developers don't rely on programming specific actions but rather specific movements.

In a TBS game the AI usually has to perform a myriad of actions that result in a certain outcome that has far more possible variations than an RTS.
For example, an AI in a 4X game attempting to come to a diplomatic resolution has to factor in an incredible amount of different factors, past actions that have been dealt against them, leverage in the form of tribute or technology, their current knowledge of the enemy's forces, and so on.
But that's only diplomacy, and even then the AI has to react practically in a way that will further its survival, it's not easy to do that. What's an AI going to do when he absolutely despises another nation but has absolutely no way of surviving a war against them? Just sit there on its ass and do nothing?

That's almost never the answer, in Civ III the AI usually did one of three things, one, it would attack the other player and thus be totally crushed, two, it would consistently ask other players to form alliances and then once they allied like one or fifty people (who could tell what the fuck they were thinking) they would attack and eventually get crushed because Civ III's AI was slow and retarded, or the third option would be is that they would constantly declare war on the hated nation, then the next turn ask the hated nation "Please spare us, here is a tribute of 400 gold." That would go on for about fifty turns until they ran out of technology and money to give. If I recall correctly Alpha Centauri's AI did this as well.
My point regarding this is that even with something as "simplistic" as diplomacy which doesn't require to factor the nearly endless amount of information a opposing nation's army make up is composed of (and other issues such as terrain, technology levels, subterfuge, etc.) and simply relies on a checklist of past transgressions and beneficial actions to dictate its disposition, the AI still tends to bug out.

And that's just one part of it, many 4X games avoid this by simplifying the diplomacy system, look at nearly any MoO game or Galactic Civilizations II, their diplomacy models don't approach anything of Civilization caliber, but eventually there are still other issues.

Let's look at ship construction for example, in MoO II GCII there are tons of possibilities, the AI can do a variety of things, they can use premade ship models (which they typically do), or they can utilize self made stuff regarding their own intuition (which they do not have), obviously the second route is going to end fucked up, so how the hell do you propose variety, something which is the essential effort of any TBS game (because watching the same thing happen slowly each turn is poor, poor gameplay) if the AI can't even manage to utilize any of this properly?
Any result would end up flawed if the AI had to create its own units because the results would almost always be off kilter in some way since the AI would construct its units (ideally, and in reality, the only sensible route) based on the army composition of other players, and as we know, that never ends up well unless the AI is astoundingly capable, which it never is.

That's why I don't think RTS AI is more difficult, sure it's real time, but it's much much simpler, and while the AI still hasn't approached anything that could be called intelligent, it's for the most part much better, Myth II's AI still impresses me to this day with its ability to properly withdraw units and even lure the player into traps at certain occasions.

And Civ IV AI cheats on the harder levels, it has the astoundingly ability of magical divination where it can perceive minute details regarding your nation that no human being could.

The only way to make a convincing AI is to develop a system of randomization that utilizes a genome (a la Creatures) like something where each game it creates a unique AI that has a construct of assembled genes from a pool of hundreds with different attributes and preferences assigned to each individual definer. Yeah, that sounds ridiculous, but it'd be the only way to create a human and intelligent opponent, one who will have various paranoia attached to it which will make it play humanly rather than mechanically.

EDIT: Wrong wording.
 
Eyenixon said:
I've never seen them have wars with each other, they typically sit on their ass and do nothing.
As I said: the traditional AI is relatively passive, but even then they do war amongst eachother. AI personality also heavily impacts this. Just look at any of the many Civ 4 game reports posted on Civ 4 websites (which are often pretty good fun to read anyway).



Eyenixon said:
The only way to make a convincing AI is to develop a system of randomization that utilizes a genome (a la Creatures) like something where each game it creates a unique AI that has a construct of assembled genes from a pool of hundreds with different attributes and preferences assigned to each individual definer. Yeah, that sounds ridiculous, but it'd be the only way to create a human and intelligent opponent, one who will have various paranoia attached to it which will make it play humanly rather than mechanically.
It is ridiculous, and I have no idea why you think that that would actually work, or be useful. You're not describing a way to create a *good* AI, you're just describing a way to create thousands of different AI personalities. That doesn't actually make them smart, and most of them will suffer from huge, huge oversights in their AI. The oversights that everyone constantly whines about when discussing normal AI.

As for the bit about real-time vs turn-based AI: you are not talking about fundamental differences that make a real-time AI easier. You're saying that creating an AI for a complex environment is harder than creating one for a relatively simple environment. That's pretty obvious, but that doesn't flow from real-time v turn-based AI, that flows from the fact that you are comparing RTS games to turn-based 4X games, 2 fundamentally different environments.

You should actually be comparing equal games. For instance, a real-time squad-based game vs a turn-based squad-based game. I'm not certain about this, as I'm not really a big fan of squad-based games, but I'm fairly certain that Jagged Alliance has a much better AI than any real-time squad-based game.

The fundamental difference is that a real-time AI is confronted with a constantly changing environment at every decision point, while a turn-based AI is confronted with a static environment. The latter is inherently easier to analyse.

Also, your diplomacy example is pretty poorly chosen as that is relatively easy to create once you've got the rest going. Diplomacy is very easily analysed and created, as there's a very well-defined realm of possibilities. That's more an example of stupid programming than difficult programming.
 
Good AI necessitates personality in the first place, you aren't going to get something that can think constructively without giving it parameters that define what it is, it's not simply an effort to make it think intelligently but to make it function intelligently as well, or rather have a natural thought process, that however artificial it may be can function beyond a simple on/off switch dictating whether or not a situation is simply dangerous or lethal and thus relegating more resources to one than the other, which seems to be the general philosophy regarding decent AI in gaming. But that really isn't good enough, the main factor is unpredictability and an AI that simply plays well is garbage and doesn't actually function as a good AI should, which is intuitively. A player who fights an AI that produces unusual strategies because of a ridiculous large set of qualifiers is going to have a much more difficult time than with an AI which, for example, counters unit composition very well but can be exploited by its failure to actually be afraid of anything and thus make counter measures. It's the difference between seeing a single air unit scouting over your base and having an AI that freaks out and starts building anti-air because it has seen the presence of an air unit, if that were to be elaborated upon, it could be said that one AI doesn't care, one AI will take greater methods towards suppressing scout patrols (sending out fast anti-light unit killers for example) and one will bug out.
I'm not saying it's fool proof, I'm definitely not saying each and every AI will be perfect and great at the game, I am saying that there should be various levels of intelligence between each difficulty. A stronger and more difficult AI will have less tendency to overreact to situations or use up all its resources on singular efforts, and it will be able to assess independently what exactly creates a larger threat based on more than "which enemy unit has killed the most of my units" or "which enemy unit is the strongest, regardless of number".

There are going to be oversights, but that's because it's a concept and not an actual construct, if it were done properly it would then require an inordinate amount of fail-safes to prevent it from doing things it shouldn't if the way the reaction is executed is free from the actual reasoning of executing a reaction. I don't think, really, that constructing an AI of this sort would come without measures to prevent certain imbecilic reactions, you know, it would also necessitate genes that allows the AI to humanly realize things that you never should do. Some AIs would be worse in this area, meaning the lower difficulty AIs. It would create more variety in difficulties.
If the AI program feels that it should mass tanks rather than another specific unit to counter infantry, then for whatever reason, that paranoia regarding infantry should be specifically tailored to the way the AI behaves in general. Which would take an insane amount of time because it would mean having all the genes interact symbiotically as they usually do in a normal healthy being, which would be insanely difficult.

That's why I added that bit as an aside, it was more or less speculation, and as speculation typically goes, it went with the assumption that my imaginative AI has an absolutely inhuman amount of effort put into it.

If you think speculation requires practicality in reference to today's level of technology, then I really think the argument is entirely irrelevant since good AI doesn't exist at all. I'm not trying to think up a practical solution, rather something that would possibly and theoretically create an AI that functions in some form as an artificial - though limited - human that provides the basic attributes of intuition and unpredictability through a large amount of specific behaviors and "genes" which dictate personal preferences towards reactions in certain situations.

Interestingly enough, your statement about real-time and turn-based squad games is hard to refute or support, simply because you've got a lot more turn-based squad games, and most of the real time ones -of which there are few- such as Commandos or Desperadoes have pretty junky AI, however, I blame that on the developers and their inability to do anything other than create unit waypoints more than the actual difficulty of programming the AI.

My diplomacy example was chosen to reveal the fact that something so inherently simple as diplomacy reveals far more faults and errors than the general execution of AI in a real-time game, call it an unequal comparison, but it reveals the faults in reasoning that are typically present in TBS games, the AI can ruin itself with a single stupid decision in diplomacy, and there isn't a single game that's done it right, this is like statistics however, the programming can be stupid if everyone has just overlooked the solution since the early 90s, but I doubt that to be the case. It's extremely difficult for the AI to perceive what will happen in the future if it isn't paranoid (in a computer-relative sense of fear) of what could possibly result. You need an AI that has expectations, however simply they may be.

I also really don't think there are any alike games in either genre, simply because the idea of them being arbitrarily different doesn't really exist. I'm going to use Fallout Tactics as an example here. If you switch to real-time mode from the various turn based ones the difference is mostly that action points are switched from a "press button to regenerate and wait" function to a "regenerate and keep playing" one, that's a pretty minuscule difference that completely changes the pacing and outcome of the game's combat, but it's also an isolated instance in that it's the same game and the same rules. You're not going to see this anywhere else as two different whole products, you're not going to see it in two separate games to be explicit, in relation to a different real-time game versus a turn-based one, that similarity doesn't exist, unless it's the same game using the same rules, like Arcanum.

I'm repeating that a lot because the most I can think of is something like Advanced Wars versus Warcraft I and that's about the level of complexity that seems equal to me, and I sincerely doubt that for the most part the genre as it exists on the PC would even consider Advanced Wars as a major contender.
I stand by my belief that TBS AI is difficult because the genre is inherently more complex. I'm not accusing you of being unpractical, but I am saying that it's practical to make that assumption because most TBS games will be far more complex because they very much produce enjoyment with variety and possibility.

Lastly, I want to raise issue with your assertion that a consistent changing environment makes it more difficult to code the AI. It does require more commands and reactions on the AI's part, but a TBS AI doesn't necessarily operate more intelligently simply because it has to make a decision at a static moment, as I've mentioned before, the various inherent complexities of a TBS game means that there's far more possibilities that could render an AI's decision stupid. You can't really blame the AI in an RTS if he sends the proper unit to counter you and then you flank him with concealed units that he couldn't see (which would require a non-cheating AI anyways), you can however blame him if he doesn't retreat or bring in new units, but a lot of RTS games address that and it's not really such a complex occurrence for the AI to react to, it'd be worse if a newer RTS didn't address this simply because it's a common expectation now, something that the AI does in Myth II from 1998.
Also it's a bit obvious that the operating of the game runs parallel to the operating of the AI but they don't always intersect, it's simple for the AI itself to have a "make this unit move" command, that's free of the unit's actual pathfinding, so I don't believe it's more difficult for the AI to make multiple units move and attack in a real-time environment than it is for an AI to make multiple decisions during a turn in a TBS game. When the AI has to react that's a whole different situation that doesn't always exist in a TBS game since a reaction is just "wait for my turn and then do something", but as I said, I believe it's more difficult for the AI to make intelligent decisions in that type of scenario than it is for it to all of a sudden retract its units using a move command if for example it feels that "large amount of units countering mine", and then sends in units to counter those that are countering his. While yes, there are far more dire and complex situations in a real-time game than that example, the programmers aren't necessarily coding like the AI can make these decisions with human speed, it can do it much faster. The difficulty is in having the AI make intelligent decisions with a large amount of "stimuli" that is occurring at a faster rate, so it has to utilize its resources properly in these situations properly by diverting them to different efforts based on necessity. Because of the speed of the game, there isn't really a need for the AI to have expectations because he is for the most part simply reacting to what he "sees", rather than having to make active decisions as in a TBS game that regards more than just his units dying or not.

Even Jagged Alliance 2's AI should have to take that into account, his unit being stranded in an unattractive position for a turn could very much be an issue.

EDIT: Oh my god.
 
Holy mother of Jebus, wall of text.

Okay, first, ad AI personalities. What you are talking about here is unpredictability, not necessarily personality differences. Unpredictability can simply be achieved through randomly choosing between strategies that are close in expected results. This doesn't necessitate the overly complicated solution you proposed.

Also, you are wrong that unpredictability is a necessity. While obscuring your strategy and not being predictable is one road to success, another is simply playing really well. See also: game theory.

You are obscuring the issue here by constructing elaborate schemes to try to recreate a human mind. A good AI for a game won't require any of that, it merely needs to produce convincing results. Similar results do not require an underlying equivalent approach.

Eyenixon said:
Interestingly enough, your statement about real-time and turn-based squad games is hard to refute or support, simply because you've got a lot more turn-based squad games, and most of the real time ones -of which there are few- such as Commandos or Desperadoes have pretty junky AI, however, I blame that on the developers and their inability to do anything other than create unit waypoints more than the actual difficulty of programming the AI.
You could do that, but I chose that simple example as it was the first to jump to mind. You could look at turn-based versus real-time RPGs as well (Temple of Elemental Evil versus Icewind Dale 2, for instance), or any other game. But comparing RTS and 4X games is truly unfair.

But it is merely an example to support the main point, that the fundamental difference between real-time and turn-based games is one of a nearly analog, constantly changing game environment and a static environment. As I said: the latter is fundamentally simpler, and hence is easier to create a good AI for.

Eyenixon said:
My diplomacy example was chosen to reveal the fact that something so inherently simple as diplomacy reveals far more faults and errors than the general execution of AI in a real-time game, called it an unequal comparison, but it reveals the faults in reasoning that are typically present in TBS games, the AI can ruin itself with a single stupid decision in diplomacy, and there isn't a single game that's done it right, this is like statistics however, the programming can be stupid if everyone has just overlooked the solution since the early 90s, but I doubt that to be the case. It's extremely difficult for the AI to perceive what will happen in the future if it isn't paranoid (in a computer-relative sense of fear) of what could possibly result. You need an AI that has expectations, however simply they may be.
It's more a case of goals and weighing values. Usually, diplomacy is fucked because it doesn't get as much attention.
But yeah, diplomacy hasn't really been done that well that often.

Eyenixon said:
I stand by my belief that TBS AI is difficult because the genre is inherently more difficult. I'm not accusing you of being unpractical, but I am saying that it's practical to make that assumption because most TBS games will be far more complex because it very much produces enjoyment with variety and possibility.
This is also part cause and effect. Designers keep real-time systems relatively simple to avoid these AI problems.

Also, your last large paragraph is actually a pretty good advocate for why real-time AI is harder to code for, as you're simply saying that a real-time AI can only quickly react and not make any long-term plans. Heh.
 
there is some good sup com FA AI packages out there

but even then they sometimes fall for obvious things.

enemy has lots of anti-air units flying around outside my base. i send out one of my T4 flying units to them. once i see on radar that their flying units are heading to my T4 unit, i have it retreat back.

they keep going after my T4 unit and ram straight in a wall of T3 anti-air units and into range of my ground based anti-air turrents. took my T4 air unit to half health and i killed like 25-30 of his T3 air units.


guess nobody told the custom-AI package what "its a trap" means :)
 
Sander said:
Holy mother of Jebus, wall of text.

Okay, first, ad AI personalities. What you are talking about here is unpredictability, not necessarily personality differences. Unpredictability can simply be achieved through randomly choosing between strategies that are close in expected results. This doesn't necessitate the overly complicated solution you proposed.

Also, you are wrong that unpredictability is a necessity. While obscuring your strategy and not being predictable is one road to success, another is simply playing really well. See also: game theory.

You are obscuring the issue here by constructing elaborate schemes to try to recreate a human mind. A good AI for a game won't require any of that, it merely needs to produce convincing results. Similar results do not require an underlying equivalent approach.

That simply doesn't work however, the results are never convincing and typically repeat, you've then got an easily exploitable AI, and simply randomizing strategies that will solve the issue is a pretty heartless solution because it just eventually creates the same effect, an AI that doesn't actually reason in a mock form of intuition, but rather just picks random numbers regardless of which is more suitable. As I've said, a good AI should have a personality to be an enjoyable adversary. I think you're mistaking what I'm saying to be an issue of whether or not the AI is simply good at the game or whether or not the AI is a good opponent.
I'm looking for accurate emulation here, not simply convincing simulation, which doesn't provide the necessary variety and depth. The minute differences a more complex system would create are far more important than their size indicates.

Sander said:
You could do that, but I chose that simple example as it was the first to jump to mind. You could look at turn-based versus real-time RPGs as well (Temple of Elemental Evil versus Icewind Dale 2, for instance), or any other game. But comparing RTS and 4X games is truly unfair.

But it is merely an example to support the main point, that the fundamental difference between real-time and turn-based games is one of a nearly analog, constantly changing game environment and a static environment. As I said: the latter is fundamentally simpler, and hence is easier to create a good AI for.

I'm talking about RTS versus TBS, it's the topic of the thread.
Anyhow, I don't believe the simplicity of the environment is as big a factor as you think it is. As the elements of a game devolve as a whole, yes, eventually a TBS game will be in fact easier to program an AI for. But when it comes down to it, as I've been saying with TBS games in general, there's simply more complexity in the systems they adapt which are harder to develop AI for because there's so much to keep track of that an AI has to respond to intelligently.

It's also an issue of time and how much effort the developer has to take. For an RTS you can develop an AI that will pretty much function on all map types, will be able to utilize all the abilities within the game, and simply use them all functionally, because there's just a lot less for the AI to have to process on a fundamental level regarding the game's systems.
That means there's a lot less to do as a whole.

We were talking about the diplomacy system for example, I think it's good that you brought up that they eventually have to develop these systems independent from each other because although these systems are different entities, they eventually effect all different aspects of the game, the AI has to predict what will happen to his economy if he adopts a new religion that requires higher upkeep but raises the happiness of his people, he has to predict what will happen to his army if he institutes a new government which promotes pacifism and bolsters the economy, he needs to know what will happen if the way he interacts with other plays during diplomacy will hurt certain factors of his empire, if trading a technology will be beneficial or strengthen a possible future enemy he's on tenuous terms with, you were saying that an RTS AI needs to predict things too, but it simply isn't as important in an RTS game, mostly because there the order of the day is react, react, react. I see this, I do this, I build my simple base to get to tech level 5 then make this sort of unit because it is stronger. Simple things, things that in a TBS game aren't so simple, that are usually far more involved and affect hundreds of other variables.

I mean, a human player acts the same way in a RTS game, he does formulate strategies and predict his enemy's movements, but for the most part he is reacting, reacting to occasions in battle or to changes in his economy which can be fixed with a simple action, rather than having to alter fifteen hundred components of his entire government, religion, and army. My point is that reactions have a lot less possibilities to account for, because they're so basic to comprehend. If you're losing, you retreat, you're getting attacked, you send units to help.
There are always more complex situations to be accounted for, but in the end, they're never as advanced as a TBS's issues where the computer has to act and not simply react.

Sander said:
This is also part cause and effect. Designers keep real-time systems relatively simple to avoid these AI problems.

Also, your last large paragraph is actually a pretty good advocate for why real-time AI is harder to code for, as you're simply saying that a real-time AI can only quickly react and not make any long-term plans. Heh.

And since the genre has evolved on that principle it's easier to say that practically it's much easier to develop an AI that functions adequately.
As I said before, reaction is the most important thing in an RTS, and since an AI can do that in an extraordinary amount of time it naturally has an advantage, its inability to plan is a definite issue, but not as major as it would be in a TBS game where that's practically the focus.
 
Eyenixon said:
That simply doesn't work however, the results are never convincing and typically repeat, you've then got an easily exploitable AI, and simply randomizing strategies that will solve the issue is a pretty heartless solution because it just eventually creates the same effect, an AI that doesn't actually reason in a mock form of intuition, but rather just picks random numbers regardless of which is more suitable. As I've said, a good AI should have a personality to be an enjoyable adversary. I think you're mistaking what I'm saying to be an issue of whether or not the AI is simply good at the game or whether or not the AI is a good opponent.
I'm looking for accurate emulation here, not simply convincing simulation, which doesn't provide the necessary variety and depth. The minute differences a more complex system would create are far more important than their size indicates.
I disagree, and I don't see how you can possibly make this claim either. You're going by a body of work that hasn't really produced a good AI ever, and then extrapolating that into a realm of possibilities that really has very little to do with that.

It's the results that matter, not the underlying concept.

And no, a static strategy does not mean that it is singularly exploitable. Again: see game theory.

Eyenixon said:
And since the genre
*smack*
Again: stop comparing RTS games to 4X games. Comparing the two makes no sense and serves no purpose.
It's a completely unfair comparison that doesn't work. You can argue that TB systems are only used for complex games, but that's a function of the game industry, not of building AIs.

Which is why I'm not responding to the rest of your post, as it is simply arguing that a 4X game is more complex than a real-time strategy game. Valid, but completely useless point.
 
Back
Top